thread 'private_repositories_synced_by_readers_once_their_syncer_lapses' panicked at crates/knotless-serve/tests/gateway.rs:3899:5:
assertion `left == right` failed
left: ["Written after the lapse", "Written while synced"]
right: ["Written after the lapse", "Written before the takedown", "Written while synced"]
It failed in the test's second half, where a member writes an issue and is then taken down on their PDS. Rarer than the others (once in four pulls' logs, in a later whole-binary pass, not labelled FLAKY). Quarantined until 2026-11-14.
How to see it
knotless run list knotless://chadtmiller.com/knotless --pull at://did:plc:bcgltzqazw5tb6k2g3ttenbj/dev.knotless.pull/3mzqnpksbaq7u --log, search for the test name.
What you expected
The same titles on every run.
Where
The runner (spaces net), main as of 2026-11-07.
Guess at the cause: "Written before the takedown" is expected in the list after the sync as it was, but the read after the takedown can be answered before the earlier sync of that issue is indexed, or the failed sync is counted as made too early. The assertion compares exact lists with no wait, so the order of the write, the sync and the read is the race.
chadtmiller.com/knotless
What happened
private_repositories_synced_by_readers_once_their_syncer_lapses(spaces net,gateway (test)) failed once on Pages are sent on as the app sends them, not once it is done, and passed in the logs of A check's log is read ahead as it is filed, and many more are kept, .../3mzqlq2co7kx3 and .../3mzqlswtk72yf:It failed in the test's second half, where a member writes an issue and is then taken down on their PDS. Rarer than the others (once in four pulls' logs, in a later whole-binary pass, not labelled FLAKY). Quarantined until 2026-11-14.
How to see it
knotless run list knotless://chadtmiller.com/knotless --pull at://did:plc:bcgltzqazw5tb6k2g3ttenbj/dev.knotless.pull/3mzqnpksbaq7u --log, search for the test name.What you expected
The same titles on every run.
Where
The runner (spaces net), main as of 2026-11-07.
Guess at the cause: "Written before the takedown" is expected in the list after the sync as it was, but the read after the takedown can be answered before the earlier sync of that issue is indexed, or the failed sync is counted as made too early. The assertion compares exact lists with no wait, so the order of the write, the sync and the read is the race.